Rate Limits

[FibreTTP]

As I understand it, requests to the Twitch API is limited to 800 per minute. How much of this does Ganymede typically use? Does that amount change during the archival of a video or a live stream?

Checking whether a channel is live every five minutes seems too infrequent, as I assume up to five minutes of the beginning of the stream will not be able to be recorded if it happens to check too late. Would lowering this interval lead to rate limiting?

Rate limits are available in response headers on every API request. Adding the remaining limit somewhere in the application (or in the logs) would make troubleshooting potentially easier.

[Zibbp]

I don't think Twitch publishes an official up-to-date number anymore, at least not on their documentation. I have not run into any rate limits. I know of users that have a 10 second check interval that don't have any issues. I recommended a minimum of 15 seconds to be safe.

The live stream check is really the only 'api intensive' part. Twitch API requests in Ganymede do basic retrying which will retry a few times before giving up.

I can look into exposing the header from API requests, maybe if it crosses a concerning threshold?

[FibreTTP]

Thanks.

I can look into exposing the header from API requests, maybe if it crosses a concerning threshold?

Maybe print a warning to the log if it uses more than 75% of the available requests before reset? (though I would also like to know the actual number all the time.) The format could be something like:

Used <requests_used> of the <request_limit> requests available ({<requests_used> / <request_limit> * 100}). Ratelimit will reset in {<now> - <reset_time>} seconds.

This would be printed each time a request is made over the threshold.